On 13/05/09, 14:21:29, Peter FELECAN <[email protected]> wrote regarding Re: 
Compiler optimization Sun Studio vs. GCC:

> >> Didn't find it either. Well, my question was motivated to really
> >> qualify
> >> the differences, lets say on a scale of 1 to 100, how do we position
> >> the
> >> 2 compilers for SPARC generated code. Saying "very bad" is relative
> >> but
> >> to what and in what measure?
> >>
> >> Finally, is that so important for the kind of packages that we
> >> provide?
> >> For 1%, 10% or 66%...
> >
> > Maybe James or some other maintainer has real, first-hand performance
> > data?

> Let James wisdom shine upon us!

What type of software, or specifically what, are we compiling?  In
many cases the "best" optimisation is "-xO3 -xspace" because fastest
CPU alone isn't best.  Unless a person is waiting or a machine is
CPU saturated we don't need to worry.  Think too about memory usage
and if it has a very short run life load times (code size), be careful
that repeated tests don't just test memory and disc cache performance.




James.



_______________________________________________
maintainers mailing list
[email protected]
https://lists.opencsw.org/mailman/listinfo/maintainers

Reply via email to